Mortgage Lenders Hand Decision-Making to AI Agents

Agentic AI changes the capacity equation in a cyclical industry. Craig Rebmann, product evangelist and managing director at Dark Matter Technologies, told the publication: "No technology provider is going to bring you volume. They're only going to bring you capacity. The lender is then responsible for determining what they're going to do with that capacity."

Frost Bank offers a concrete example. The Texas lender re-entered mortgage lending after exiting more than 20 years earlier. It ended 2025 with $595 million in unpaid mortgage balance from originations, exceeding its goals by 19%. The bank used AI to handle the workload without a large hiring push.

The most consequential use of AI lending may be happening in markets where traditional credit infrastructure barely exists. Ahmed Mohsen, co-founder and CTO of MNT-Halan, wrote in a World Economic Forum article that Egypt's fastest-growing FinTech built an AI-powered alternative credit scoring engine that evaluates users who have never interacted with formal credit systems. MNT-Halan has automated more than 50% of its loan approvals and achieved a 60% approval rate for previously unscoreable users. The engine draws on data from the Halan superapp across payments, savings, and e-commerce.

The pattern is not limited to the U.S. or Egypt. The stock market analysis of mortgage technology companies shows growing investor interest in firms that can automate credit decisions. The Tavant deployment shows the path: provisioned agents with audit trails. MNT-Halan has automated more than 50% of its loan approvals and achieved a 60% approval rate for previously unscoreable users, Mohsen wrote.